There are many situations which require rotation of an optical waveguide, specifically an optical fiber about its longitudinal axis. For example, it may be necessary to control, statically or dynamically, e.g., by applying modulation, the state of polarization (SOP) of guided light propagating along the fiber. More particularly, it is sometimes desirable to ensure that a localized birefringence element within the fiber path is optimally aligned with the SOP of the light incident upon it, this in order to effect a transformation of the incident SOP into a substantially different output SOP. It may be desirable to effect one or several such consecutive transformations of the guided-light SOP to randomly vary the SOP of light being used to make measurements, for example to determine polarization mode dispersion (PMD) or polarization-dependent loss (PDL), when it may be desirable to select many different SOPs or even “scramble” the SOP to distribute the SOP, preferably uniformly, around the Poincaré sphere.
In the context of this specification, the verb “control” signifies “to deliberately change or maintain”. It does not necessarily entail a knowledge of the specific SOP of the output light.
For use in many commercial applications, such as the testing of links in optical networks, it is desirable for a polarization controller to exhibit or have (i) low PMD, (ii) low insertion loss; (iii) low activation loss; (iv) high power handling capacity; (v) low component cost; (vi) ease of assembly and calibration; (vii) small volume; (viii) low electrical power consumption; (ix) the ability to maintain a fixed SOP stable over a relatively long period of time, for example one hour, where static SOP settings are involved; (x) the ability to stop scrambling abruptly where a particular, generally unknown and randomly chosen SOP is required, for example during PMD or PDL measurements; and (xi) optical path integrity, including having no deleterious effect on the fiber mechanical integrity and having no effect on insertion loss if electrical power is lost.
Although bulk-optics retardance elements, for example a series of half- and quarter-waveplates, can be used to adjust SOP, they are not particularly compact or robust, and have a relatively high insertion loss and handle relatively low power since the light beam must exit the fiber and be collimated for incidence upon the waveplate surfaces, and subsequently be re-injected into a downstream fiber. Also, the size of the bulk-optic elements makes them generally unsuitable for hand-held optical test instruments.
Known polarization controllers which do not use bulk optics may fulfill one or more of requirements (i) to (xi), but not all. Thus, it is known to use one or more variable retardation elements, for example liquid crystal devices, as disclosed in U.S. Pat. No. 4,979,235 and U.S. Pat. No. 7,085,052, but they disadvantageously do not fulfill requirements (i) to (vi) and, in some cases, requirement (viii).
U.S. Pat. No. 6,973,247 discloses a polarization controller in which an optical fiber is heated by means of an internal electrode so that expansion causes compression forces which change the birefringence of the fiber. While such a controller might satisfy many of requirements (i) to (x) it is unlikely to provide long-term stability or be small and inexpensive to produce. Certainly, it would not be able to stop scrambling abruptly.
Polarization controllers also are known which bend an optical fiber to induce birefringence and then rotate the bent fiber, thereby controlling the SOP of light propagating in the optical fiber, as disclosed, for example, in U.S. Pat. No. 4,389,090 and U.S. Pat. No. 4,793,678. Disadvantageously, such devices usually require a relatively large volume because of the length of the optical fiber and its limited bending radius, and, even if motorized, would operate relatively slowly.
It is also known to apply a variable transverse compressive force to a length of optical fiber so as introduce stress-induced birefringence photo-elastically, thereby controlling SOP of light propagating along the fiber. Examples are disclosed in U.S. Pat. Nos. 6,480,637, 6,493,474, 6,754,404 and 6,873,783, all by X. Steve Yao, who explains that applying such transverse compressive force, induced electrically by a piezo-electric actuator (PEA) or the like, to different portions of the fiber by means of several “fiber squeezers” which have different orientations and apply different compressive forces, allows the SOP of the light in the fiber to be rotated about orthogonal axes on the Poincaré sphere. Drawbacks of such a PEA “fiber squeezer”-based polarization controller, however, include relatively high PMD, since the applied birefringence corresponds to a higher-order waveplate so as to operate in the approximately linear displacement regime of the PEA, inability to maintain a fixed SOP stable over a lengthy period of time, relatively large volume and relatively difficult assembly and calibration. Also, relatively high voltages need be applied to piezo-electric crystals, which might be hazardous in explosive environments, for instance, and the high fiber stress levels that need be applied for higher-order waveplate behavior have implications for optical path integrity.
In his U.S. Pat. No. 5,561,726, X. Steve Yao discloses a polarization controlling apparatus comprising two spaced anchorages for securing opposite distal portions of the fiber to prevent their rotation. A rotatable fiber squeezer positioned between the two anchorages applies compressive force to the medial portion of the fiber to change its birefringence. The fiber-squeezer itself is rotated about the longitudinal axis of the fiber so as to twist it relative to the anchorages, thereby to adjust the SOP of light propagating along the fiber.
Although this apparatus may be well suited for the primary application described in U.S. Pat. No. 5,561,726, namely polarization state “adjustment”, it is not entirely satisfactory or really suitable for use as a polarization scrambler, where uniform mapping of a, generally arbitrary, SOP onto the Poincaré sphere is required. For instance, for a certain arbitrary input SOP, it may be very difficult to select a desired output SOP (i.e., very sensitive to the rotation and/or pressure adjustment), whereas for the same input SOP, it may be very easy to select a different SOP (i.e., the sensitivity of the rotation and/or pressure adjustment can vary strongly with desired output SOP). For this reason, this device would not be very suitable for adaptation (e.g., by replacing the pressure screw of the fiber-squeezer with a PEA) so that it functioned as a random polarization scrambler that uniformly covered the Poincaré sphere. Moreover, his polarization controller design would not be entirely satisfactory for incorporating into a hand-held test instrument because the diameter of the rotatable fiber-squeezer (including the PEA) would be relatively large.
A further disadvantage of compressing portions of the fiber by means of a PEA is that it may increase the risk of fracturing and breaking of the fiber, since, in order to maintain this PEA in the linear response regime, the applied stress on the fiber induces a birefringence that corresponds to a high-order waveplate, i.e., more than the 0 to λ retardance range necessary for a basic polarization controller or scrambler. In U.S. Pat. No. 6,493,474, Yao discusses prior U.S. Pat. No. 4,988,169 (Walker), U.S. Pat. No. 4,753,507 (De Paula et al.) and U.S. Pat. No. 5,903,684 (Payton) and explains that problems arise because the applied pressure causes fracturing and breakage of the fiber. Yao decries attempts to solve the problem by coating the fiber with metal prior to applying pressure on the grounds that uniform metal coatings are not easily reproducible in production. Yao proposes instead polishing the fiber-squeezing surfaces to reduce irregularities to less than 100 microns and providing a polyimide coating between the fiber core and the surface upon which the fiber squeezers act. This approach also is not entirely satisfactory, however, because such polishing is time consuming and costly, thus in contradiction with either or both of requirements (v) or (vi) enumerated hereinbefore.